WO2024047748A1 - Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo - Google Patents

Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo Download PDF

Info

Publication number
WO2024047748A1
WO2024047748A1 PCT/JP2022/032586 JP2022032586W WO2024047748A1 WO 2024047748 A1 WO2024047748 A1 WO 2024047748A1 JP 2022032586 W JP2022032586 W JP 2022032586W WO 2024047748 A1 WO2024047748 A1 WO 2024047748A1
Authority
WO
WIPO (PCT)
Prior art keywords
behavior
recognition
gaze target
gaze
video
Prior art date
Application number
PCT/JP2022/032586
Other languages
English (en)
Japanese (ja)
Inventor
康敬 馬場崎
勝彦 高橋
隆平 安藤
浩一 二瓶
フロリアン バイエ
孝法 岩井
勇人 逸身
Original Assignee
日本電気株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 日本電気株式会社 filed Critical 日本電気株式会社
Priority to PCT/JP2022/032586 priority Critical patent/WO2024047748A1/fr
Publication of WO2024047748A1 publication Critical patent/WO2024047748A1/fr

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/167—Position within a video image, e.g. region of interest [ROI]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00—Television systems
    • H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Definitions

  • the present disclosure relates to a video processing system, a video processing method, and a video processing device.
  • Patent Document 1 is known as a related technology.
  • Patent Document 1 describes a remote monitoring system that transmits images captured by a camera mounted on a vehicle via a network and analyzes the images at a remote monitoring center. It is described that the image quality is improved and other areas are sent with lower image quality.
  • the present disclosure aims to provide a video processing system, a video processing method, and a video processing device that can appropriately control the amount of video data.
  • a video processing system includes an image quality control unit that controls the image quality of a gaze area including a gaze target in an input video, and a recognition unit that performs recognition processing regarding an object included in the video whose image quality of the gaze area is controlled. and an extraction means for extracting the gaze target based on the uncertainty of the recognition result of the recognition process.
  • a video processing method controls the image quality of a gaze region including a gaze target in an input video, performs recognition processing regarding an object included in the video whose image quality of the gaze region is controlled, and performs recognition processing on an object included in the video for which the image quality of the gaze region is controlled.
  • the object of attention is extracted based on the indeterminacy of the recognition result.
  • a video processing device includes an image quality control unit that controls the image quality of a gaze area including a gaze target in an input video, and a recognition process that performs a recognition process regarding an object included in the video whose image quality of the gaze area is controlled. and an extraction means for extracting the gaze target based on the uncertainty of the recognition result of the recognition process.
  • FIG. 1 is a configuration diagram showing an overview of a video processing system according to an embodiment.
  • FIG. 1 is a configuration diagram showing an overview of a video processing device according to an embodiment.
  • 1 is a flowchart showing an overview of a video processing method according to an embodiment.
  • FIG. 1 is a configuration diagram showing the basic configuration of a remote monitoring system.
  • 1 is a configuration diagram showing a configuration example of a terminal according to Embodiment 1.
  • FIG. 1 is a configuration diagram showing an example configuration of a center server according to Embodiment 1.
  • FIG. FIG. 2 is a configuration diagram showing a configuration example of a behavior recognition unit according to Embodiment 1.
  • FIG. FIG. 2 is a configuration diagram showing a configuration example of a predictor according to Embodiment 1.
  • FIG. 1 is a configuration diagram showing an overview of a video processing system according to an embodiment.
  • FIG. 1 is a configuration diagram showing an overview of a video processing device according to an embodiment.
  • 1 is a flow
  • FIG. 3 is a flowchart illustrating an example of the operation of the remote monitoring system according to the first embodiment.
  • FIG. 3 is a diagram for explaining video acquisition processing according to the first embodiment.
  • FIG. 3 is a diagram for explaining object detection processing according to the first embodiment.
  • 5 is a flowchart illustrating an operation example of behavior recognition processing according to the first embodiment.
  • FIG. 3 is a diagram for explaining behavior recognition processing according to the first embodiment.
  • 5 is a flowchart illustrating an operation example of gaze target extraction processing according to the first embodiment.
  • FIG. 7 is a diagram for explaining an example in which there are variations in recognition results in the gaze target extraction process according to the first embodiment.
  • FIG. 7 is a diagram for explaining another example in which there are variations in recognition results in the gaze target extraction process according to the first embodiment.
  • FIG. 3 is a diagram for explaining an example in which there is no variation in recognition results in the gaze target extraction process according to the first embodiment; 7 is a diagram for explaining another example in which there is no variation in recognition results in the gaze target extraction process according to the first embodiment; FIG. FIG. 3 is a diagram for explaining gaze area determination processing according to the first embodiment.
  • 7 is a configuration diagram showing a configuration example of a behavior recognition unit according to Embodiment 2.
  • FIG. 7 is a flowchart illustrating an operation example of behavior recognition processing according to Embodiment 2.
  • FIG. FIG. 7 is a diagram for explaining dropout processing according to Embodiment 2; 7 is a diagram for explaining dropout processing according to Embodiment 2.
  • FIG. FIG. 1 is a configuration diagram showing an overview of the hardware of a computer according to an embodiment.
  • the bandwidth of the network that transmits the video is limited, so it is preferable to suppress the amount of video data to be transmitted as much as possible.
  • the amount of video data can be reduced by increasing the video compression rate.
  • the video compression rate is high or the data loss rate is high, the number of erroneous recognitions increases and the recognition accuracy decreases. Therefore, in the embodiment, it is possible to prevent erroneous recognition while suppressing the amount of video data to be transmitted as much as possible. For example, it prevents misrecognition of important events such as unsafe or dangerous actions.
  • FIG. 1 shows a schematic configuration of a video processing system 10 according to an embodiment.
  • the video processing system 10 is applicable to, for example, a remote monitoring system that collects video via a network and monitors the video.
  • the video processing system 10 includes an image quality control section 11, a recognition section 12, and an extraction section 13.
  • the image quality control unit 11 controls the image quality of the gaze area including the gaze target in the input video. For example, the image quality control unit 11 may make the image quality of the gaze area higher than that of other areas, that is, make it clearer.
  • the recognition unit 12 performs recognition processing regarding an object included in a video whose image quality in a gaze area is controlled.
  • the object-related recognition process is an action recognition process that recognizes the behavior of the object, but it may also be a process that recognizes information or characteristics regarding other objects.
  • the extraction unit 13 extracts the gaze target based on the uncertainty of the recognition result of the recognition process by the recognition unit 12.
  • Indeterminacy of recognition results refers to variations in recognition results.
  • the recognition unit 12 includes a plurality of behavior predictors that have learned different learning data, and the extraction unit 13 determines the gaze target based on variations in recognition results of a plurality of behaviors output from the plurality of behavior predictors. May be extracted.
  • the gaze target may be extracted based on the behavior label included in the behavior recognition result or the variation in the score of the behavior label.
  • the recognition unit 12 includes one behavior predictor, and the extraction unit 13 extracts the information based on the dispersion of the recognition results of multiple behaviors output when the one behavior predictor performs behavior recognition multiple times.
  • a gaze target may be extracted. Extracting a gaze target means selecting a gaze target from candidates included in the recognition result.
  • FIG. 2 shows the configuration of a video processing device 20 according to the embodiment.
  • the video processing device 20 may include the image quality control section 11, the recognition section 12, and the extraction section 13 shown in FIG.
  • part or all of the video processing system 10 may be placed at the edge or in the cloud.
  • the edge is a device placed at or near the site, and is also a device close to the terminal as a layer of the network.
  • the image quality control unit 11 may be placed in an edge terminal, and the recognition unit 12 and extraction unit 13 may be placed in a cloud server.
  • each function may be distributed and arranged in the cloud.
  • FIG. 3 shows a video processing method according to an embodiment.
  • the video processing method according to the embodiment is executed by the video processing system 10 in FIG. 1 or the video processing device 20 in FIG. 2.
  • the image quality of the gaze area including the gaze target in the input video is controlled (S11).
  • a recognition process is performed regarding an object included in the video whose image quality in the gaze area has been controlled (S12).
  • a gaze target is extracted based on the indeterminacy of the recognition result of the recognition process (S13).
  • the image quality of the gaze area including the extracted gaze target is controlled for the input video.
  • the video processing system recognizes the behavior of an object from a video whose image quality has been controlled, and extracts a gaze target based on the indeterminacy of the recognition result. For example, if the recognition results vary, actions may not be recognized correctly, and the reliability of the recognition results may be low. Therefore, by making the object corresponding to the recognition result the object of attention and making it clearer, actions and the like can be recognized correctly. Furthermore, since areas other than the target to be watched can be compressed, the amount of video data to be transmitted can be suppressed.
  • FIG. 4 shows the basic configuration of the remote monitoring system 1.
  • the remote monitoring system 1 is a system that monitors an area where images are taken by a camera.
  • the system will be described as a system for remotely monitoring the work of workers at the site.
  • the site may be an area where people and machines operate, such as a work site such as a construction site, a public square where people gather, or a school.
  • the work will be described as construction work, civil engineering work, etc., but is not limited thereto.
  • the remote monitoring system can be said to be a video processing system that processes videos, and also an image processing system that processes images.
  • the remote monitoring system 1 includes a plurality of terminals 100, a center server 200, a base station 300, and an MEC 400.
  • the terminal 100, base station 300, and MEC 400 are placed on the field side, and the center server 200 is placed on the center side.
  • the center server 200 is located at a data center, monitoring center, or the like that is located away from the site.
  • the field side is the edge side of the system, and the center side is also the cloud side.
  • the center server 200 may be composed of one device or may be composed of a plurality of devices. Further, part or all of the center server 200 may be placed in the cloud.
  • the video recognition function 201 and the alert generation function 202 may be placed in the cloud
  • the GUI drawing function 203 and the screen display function 204 may be placed in a monitoring center or the like.
  • Terminal 100 and base station 300 are communicably connected via network NW1.
  • the network NW1 is, for example, a wireless network such as 4G, local 5G/5G, LTE (Long Term Evolution), or wireless LAN.
  • Base station 300 and center server 200 are communicably connected via network NW2.
  • the network NW2 includes, for example, core networks such as 5GC (5th Generation Core network) and EPC (Evolved Packet Core), the Internet, and the like. It can also be said that the terminal 100 and the center server 200 are communicably connected via the base station 300.
  • 5GC Fifth Generation Core network
  • EPC Evolved Packet Core
  • the base station 300 and MEC 400 are communicably connected by any communication method, the base station 300 and MEC 400 may be one device.
  • the terminal 100 is a terminal device connected to the network NW1, and is also a video generation device that generates on-site video.
  • the terminal 100 acquires an image captured by a camera 101 installed at the site, and transmits the acquired image to the center server 200 via the base station 300.
  • the camera 101 may be placed outside the terminal 100 or inside the terminal 100.
  • the terminal 100 compresses the video from the camera 101 to a predetermined bit rate and transmits the compressed video.
  • the terminal 100 has a compression efficiency optimization function 102 that optimizes compression efficiency and a video transmission function 103 .
  • the compression efficiency optimization function 102 performs ROI control to control the image quality of a ROI (Region of Interest).
  • the compression efficiency optimization function 102 reduces the bit rate by lowering the image quality of the region around the ROI while maintaining the image quality of the ROI including the person or object.
  • the video transmission function 103 transmits the quality-controlled video to the center server 200.
  • the base station 300 is a base station device of the network NW1, and is also a relay device that relays communication between the terminal 100 and the center server 200.
  • the base station 300 is a local 5G base station, a 5G gNB (next Generation Node B), an LTE eNB (evolved Node B), a wireless LAN access point, or the like, but may also be another relay device.
  • MEC 400 is an edge processing device placed on the edge side of the system.
  • the MEC 400 is an edge server that controls the terminal 100, and has a compression bit rate control function 401 and a terminal control function 402 that control the bit rate of the terminal.
  • the compression bit rate control function 401 controls the bit rate of the terminal 100 through adaptive video distribution control and QoE (quality of experience) control. For example, the compression bit rate control function 401 predicts the recognition accuracy that will be obtained while suppressing the bit rate according to the communication environment of the networks NW1 and NW2, and sets the bit rate to the camera 101 of each terminal 100 so as to improve the recognition accuracy. Assign.
  • the terminal control function 402 controls the terminal 100 to transmit video at the assigned bit rate. Terminal 100 encodes the video at the allocated bit rate and transmits the encoded video.
  • the center server 200 is a server installed on the center side of the system.
  • the center server 200 may be one or more physical servers, or may be a cloud server built on the cloud or other virtualized servers.
  • the center server 200 is a monitoring device that monitors on-site work by recognizing people's work from on-site camera images.
  • the center server 200 is also a recognition device that recognizes the actions of people in the video transmitted from the terminal 100.
  • the center server 200 has a video recognition function 201, an alert generation function 202, a GUI drawing function 203, and a screen display function 204.
  • the video recognition function 201 inputs the video transmitted from the terminal 100 into a video recognition AI (Artificial Intelligence) engine to recognize the type of work performed by the worker, that is, the type of behavior of the person.
  • the alert generation function 202 generates an alert in response to the recognized work.
  • the GUI drawing function 203 displays a GUI (Graphical User Interface) on the screen of a display device.
  • the screen display function 204 displays images of the terminal 100, recognition results, alerts, etc. on the GUI.
  • Embodiment 1 Next, Embodiment 1 will be described. In this embodiment, an example will be described in which a gaze target is extracted based on variations in action recognition results obtained by a plurality of predictors.
  • FIG. 4 shows a configuration example of the terminal 100 and the center server 200
  • FIG. 5 shows a configuration example of terminal 100 according to this embodiment
  • FIG. 6 shows a configuration example of center server 200 according to this embodiment.
  • the configuration of each device is an example, and other configurations may be used as long as the operation according to the present embodiment described later is possible.
  • some functions of the terminal 100 may be placed in the center server 200 or other devices, or some functions of the center server 200 may be placed in the terminal 100 or other devices.
  • the terminal 100 includes a video acquisition section 110, a detection section 120, an image quality change determination section 130, a compression efficiency determination section 140, and a terminal communication section 150.
  • the video acquisition unit 110 acquires the video captured by the camera 101 (also referred to as input video).
  • the input video includes a person who is a worker working on a site, a work object used by the person, and the like.
  • the video acquisition unit 110 is also an image acquisition unit that acquires a plurality of time-series images.
  • the detection unit 120 is an object detection unit that detects an object within the acquired input video.
  • the detection unit 120 detects an object in each image included in the input video, and assigns a label to the detected object, that is, an object label.
  • the object label is the class of the object and indicates the type of the object. For example, object labels include people, cars, robots, hammers, etc.
  • the detection unit 120 extracts a rectangular region containing an object from each image included in the input video, recognizes the object within the extracted rectangular region, and assigns a label to the recognized object.
  • the rectangular area is a bounding box or an object area. Note that the object area including the object is not limited to a rectangular area, but may be a circular area, an irregularly shaped silhouette area, or the like.
  • the detection unit 120 calculates the feature amount of the image of the object included in the rectangular area, and recognizes the object based on the calculated feature amount. For example, the detection unit 120 recognizes objects in an image using an object recognition engine that uses machine learning such as deep learning. Objects can be recognized by machine learning the features of the object image and the object label.
  • the object detection result includes an object label, position information of a rectangular area including the object, and the like.
  • the position information of the object is, for example, the coordinates of each vertex of a rectangular area, but it may also be the position of the center of the rectangular area, or the position of any point on the object.
  • the detection unit 120 transmits the detection result of the object to the image quality change determination unit 130.
  • the image quality change determination unit 130 determines a region of interest (ROI) that is an image quality change area in which the image quality of the acquired input video is changed.
  • ROI region of interest
  • the image quality change determination unit 130 is a determination unit that determines the gaze area.
  • the gaze area is an area that includes the gaze target, and is an area where the image quality is increased, that is, the image quality is made clearer. Furthermore, the gaze area can be said to be an area that ensures image quality for action recognition.
  • the image quality change determination unit 130 includes a first determination unit 131 and a second determination unit 132.
  • first the first determination unit 131 determines the gaze area
  • the second determination unit 132 determines the gaze area. Note that the determination of the gaze area by the first determination unit 131 may be omitted, and only the determination of the gaze area by the second determination unit 132 may be performed.
  • the first determination unit 131 determines the gaze area of the input video based on the detection result of the object detected within the input video.
  • the first determination unit 131 determines a gaze area based on position information of an object having a label to be gazed among detection objects detected in the input video of the detection unit 120.
  • the gaze target may be a person who is the target of behavior recognition, or may include a work object that the person can use in work.
  • the label of a work object is preset as a label of an object related to a person.
  • the target of action recognition is not limited to a person, but may also be an object such as heavy machinery or a robot. That is, actions including work performed by heavy machinery or robots may be recognized.
  • the second determination unit 132 determines the gaze area of the input video based on the fed back information.
  • extracted gaze target information which is information on the gaze target extracted by the center server 200
  • the extracted gaze target information is information regarding the gaze target, and is information indicating the gaze target extracted by the center server 200 performing behavior recognition.
  • the extracted gaze target information is position information of the gaze target, and includes position information of a rectangular region of the gaze target.
  • the second determination unit 132 determines the rectangular area indicated by the acquired extracted gaze target information as the gaze area. That is, an area that ensures the image quality of the input video is determined based on the extracted position of the gaze target.
  • the compression efficiency determining unit 140 determines the compression rate of the region of interest or an area other than the region of interest, and compresses the video.
  • the compression efficiency determining unit 140 is an encoder that encodes the input video using the determined compression rate.
  • the compression efficiency determination unit 140 may be configured, for example, by H. 264 and H.
  • the video is encoded using a video encoding method such as H.265.
  • the compression efficiency determining unit 140 encodes the input video so that the bit rate assigned by the compression bit rate control function 401 of the MEC 400 is achieved.
  • the compression efficiency determination unit 140 is an image quality control unit that controls the image quality of the gaze area determined by the image quality change determination unit 130, and corresponds to the image quality control unit 11 in FIG. It can also be said that the compression efficiency determination unit 140 is an image quality improvement unit that improves the image quality of the region of interest.
  • the gaze area is an area determined by either the first determination unit 131 or the second determination unit 132.
  • the compression efficiency determining unit 140 encodes the image quality of the image area to a predetermined quality by compressing the image area and other areas at predetermined compression rates. That is, by changing the compression ratio between the focused area and other areas, the image quality of the focused area is made higher than that of the other areas. It can also be said that the image quality of other areas is lower than that of the gaze area.
  • the image quality of the gaze area and other areas is controlled within the bit rate assigned by the compression bit rate control function 401 of the MEC 400.
  • the image quality of the gaze area may be controlled by changing not only the compression rate but also the image resolution, frame rate, and the like.
  • the image quality of the gaze area may be controlled by changing the amount of color information of the image, for example, color, gray scale, black and white, etc.
  • the terminal communication unit 150 transmits the encoded data encoded by the compression efficiency determination unit 140 to the center server 200 via the base station 300.
  • the terminal communication unit 150 is a transmitting unit that transmits a video whose image quality in the viewing area is controlled. Furthermore, the terminal communication unit 150 receives extracted gaze target information transmitted from the center server 200 via the base station 300.
  • the terminal communication unit 150 is an acquisition unit that acquires extracted gaze target information.
  • the terminal communication unit 150 is an interface that can communicate with the base station 300, and is, for example, a wireless interface such as 4G, local 5G/5G, LTE, or wireless LAN, but may also be a wireless or wired interface of any other communication method. good.
  • the terminal communication unit 150 may include a first terminal communication unit that transmits encoded data and a second terminal communication unit that receives extracted gaze target information.
  • the first terminal communication section and the second terminal communication section may be communication sections using the same communication method, or may be communication sections using different communication methods.
  • the center server 200 includes a center communication section 210, a decoder 220, an action recognition section 230, an analysis information storage section 240, and a gaze target analysis section 250.
  • the center communication unit 210 receives encoded data transmitted from the terminal 100 via the base station 300.
  • the center communication unit 210 is a receiving unit that receives video whose image quality in the viewing area is controlled. Further, the center communication unit 210 transmits the extracted gaze target information extracted by the gaze target analysis unit 250 to the terminal 100 via the base station 300.
  • the center communication unit 210 is a notification unit that notifies the extracted gaze target information.
  • the center communication unit 210 is an interface capable of communicating with the Internet or a core network, and is, for example, a wired interface for IP communication, but may be a wired or wireless interface of any other communication method.
  • the center communication unit 210 may include a first center communication unit that receives encoded data and a second center communication unit that transmits extracted gaze target information.
  • the first center communication section and the second center communication section may be communication sections using the same communication method, or may be communication sections using different communication methods.
  • the decoder 220 decodes the encoded data received from the terminal 100.
  • the decoder 220 corresponds to the encoding method of the terminal 100, for example, H. 264 and H.
  • the video is decoded using a video encoding method such as H.265.
  • the decoder 220 decodes each area according to the compression rate and generates a decoded video (also referred to as received video).
  • the behavior recognition unit 230 is a recognition unit that recognizes the behavior of an object in the decoded received video, and corresponds to the recognition unit 12 in FIG. 1.
  • the behavior recognition unit 230 executes behavior recognition processing for recognizing the behavior of the gaze target on the video whose image quality in the gaze area is controlled.
  • the action recognition unit 230 detects an object from the received video and recognizes the action of the detected object.
  • the behavior recognition unit 230 recognizes the behavior of the person who is the target of behavior recognition, and assigns a label of the recognized behavior, that is, a behavior label.
  • the behavior label is a class of behavior and indicates the type of behavior.
  • the behavior recognition unit 230 recognizes the behavior of a person based on the person and the work object detected from the received video.
  • the behavior recognition unit 230 may recognize the behavior of a person by identifying the relationship between the person and the work object.
  • the relationship between a person and a work object includes which object the person is using or not using.
  • the work object may be identified for each person based on the distance between the person and the work object, and the behavior may be recognized from the identified work object.
  • the behavior recognition unit 230 performs machine learning on work objects and tasks related to a person, and recognizes the behavior of the person based on machine learning.
  • the method is not limited to the machine learning basis, and may also be used to associate work objects and tasks related to a person and recognize the person's actions based on rules.
  • a work object and a work content may be associated in advance, and a person's behavior may be recognized based on the detected work object.
  • actions may be recognized only from the person.
  • the posture and shape of the person may be associated with the content of the work in advance, and the behavior of the person may be recognized based on the detected posture and shape of the person.
  • the action recognition unit 230 includes a plurality of predictors that predict actions from received videos, and outputs action recognition results predicted by the plurality of predictors.
  • the analysis information storage unit 240 stores analysis information analyzed by the behavior recognition unit 230.
  • the analysis information includes action recognition results, person detection information, work object detection information related to the action, and the like.
  • the action recognition result may include detection information of a person and detection information of a work object related to the action.
  • the action recognition result includes a label of the recognized action, a score of the action label, identification information of the person performing the recognized action, identification information of the work object used in the recognized action, and the like.
  • the score of the behavior label indicates the degree of certainty, which is the probability (probability) of the behavior label. The higher the score, the more likely the predicted behavior label is correct.
  • the person detection information includes position information of a rectangular area of the person, tracking information, and the like.
  • the tracking information is trajectory information indicating the tracking result of the object.
  • the detection information of the work object includes an object label, a score of the object label, position information of a rectangular area of the object, tracking information, and the like.
  • the behavior predictor (behavior recognition engine) of the behavior recognition unit 230 extracts candidates for work objects that can be related to each image by learning to give weight to objects related to the action, and Outputs information about object candidates. For example, when it recognizes a pile-driving operation, it outputs information about a hammer, which is an object related to the action.
  • the gaze target analysis unit 250 is an extraction unit that extracts a gaze target based on the analysis information analyzed by the behavior recognition unit 230, and corresponds to the extraction unit 13 in FIG. 1.
  • the analysis information may be acquired from the behavior recognition section 230 or from the analysis information storage section 240.
  • the gaze target analysis unit 250 determines a gaze target that ensures image quality in order to prevent behavioral recognition errors.
  • the gaze target analysis unit 250 determines the gaze target based on the action recognition result.
  • the gaze target analysis unit 250 targets a person whose behavior is recognized by the behavior recognition unit 230, that is, a person whose behavior is included in the behavior recognition result.
  • the person and the work object may be set as the gaze targets.
  • objects related to the work may be a "pile” and a "hammer", and the person, the "pile” and the “hammer” may be set as objects of attention.
  • the gaze target analysis unit 250 extracts the gaze target based on the indeterminacy of the action recognition result.
  • the gaze target is determined based on variations in the plurality of action recognition results output by the plurality of predictors of the action recognition unit 230, respectively.
  • the target of attention is determined based on the variation in behavior labels and the variation in scores of behavior labels included in multiple behavior recognition results. For example, if the dispersion of the action recognition results is larger than a predetermined range, objects including the person who performed the action of the action label and the work object may be determined as the gaze target.
  • the gaze target analysis unit 250 outputs the position information of the extracted rectangular region of the gaze target as extracted gaze target information.
  • the position information is, for example, the coordinates of each vertex of the rectangular area, but may also be the position of the center of the rectangular area, or the position of any point of the gaze target.
  • the extracted gaze target information includes not only location information but also information analyzed by the behavior recognition unit 230, such as object labels and image features of the gaze target, behavior labels, and behavior label scores, as information regarding the extracted gaze target. But that's fine.
  • FIG. 7 shows a configuration example of the behavior recognition unit 230 in the center server 200.
  • the behavior recognition unit 230 includes a plurality of predictors PM1 to PM3.
  • the number of predictors PM is not limited to three, and any number of predictors PM may be provided.
  • Predictors PM1 to PM3 each predict the behavior of the object in the received video, that is, recognize the behavior.
  • the predictors PM1 to PM3 may be learning models having the same configuration, but are learning data sets of different learning data. For example, the predictors PM1 to PM3 learn from videos taken of the same action or the same type of action at different work sites (environments) as learning data.
  • the learning model of the predictor is an object recognition engine or an action recognition engine.
  • the predictors PM1 to PM3 may be learning models that have learned different behaviors or different types of behaviors. For example, one predictor may learn a first behavior, such as behavior in a digging operation, and another predictor may learn a second behavior, such as behavior in a grading operation.
  • the learning data for the excavation work and the learning data for the land leveling work may be videos taken in the same environment.
  • the different actions learned by the predictors may be actions that can be performed simultaneously or actions that cannot be performed simultaneously.
  • FIG. 8 shows a configuration example of the predictor PM in FIG. 7.
  • FIG. 8 is a configuration example in which behavior recognition based on the relationship between a person and a work object is performed based on machine learning.
  • the predictor PM of the behavior recognition unit 230 includes an object detection unit 231, a tracking unit 232, a behavior predictor 233, and a behavior determination unit 234.
  • the object detection unit 231 detects an object in the input received video.
  • the object detection unit 231 is a detection unit such as an object recognition engine using machine learning. That is, the object detection unit 231 extracts a rectangular area containing an object from each image of the received video, recognizes the object within the extracted rectangular area, and assigns a label to the recognized object.
  • the object detection result includes an object label and position information of a rectangular area containing the object.
  • the tracking unit 232 tracks the detected object in the received video.
  • the tracking unit 232 associates objects in each image included in the received video based on the object detection results. By assigning a tracking ID to a detected object, each object can be identified and tracked. For example, by matching objects between images based on the distance or overlap (for example, IoU: Intersection over Union) between the rectangular area of the object detected in the previous image and the rectangular area of the object detected in the next image, Track objects.
  • IoU Intersection over Union
  • the behavior predictor 233 predicts the behavior of each object tracked by the tracking unit 232.
  • the behavior predictor 233 recognizes the behavior of the person tracked within the received video and assigns a label of the recognized behavior.
  • the behavior predictor 233 recognizes the behavior of a person in the received video using a behavior recognition engine that uses machine learning such as deep learning.
  • the behavior of a person can be recognized by machine learning of the video of the person performing the work using the work object and the behavior label. For example, learning data that is a video of a person performing a task using a work object, annotation information such as the position of the person and work object, and related information between the person and the object, and behavioral information such as the work object necessary for each task. Machine learning using .
  • the behavior predictor 233 outputs the score of the recognized behavior label.
  • the behavior determination unit 234 determines the behavior of the object based on the predicted behavior label.
  • the behavior determination unit 234 determines the behavior of the person based on the score of the behavior label predicted by the behavior predictor 233. For example, the behavior determination unit 234 outputs the behavior label with the highest score as the recognition result.
  • the recognition result may include scores of a plurality of behavior labels predicted by the behavior predictor 233.
  • FIG. 9 shows an example of the operation of the remote monitoring system 1.
  • the terminal 100 executes S101 to S105 and S111 to S112 and the center server 200 executes S106 to S110
  • the present invention is not limited to this, and any device may execute each process.
  • the terminal 100 acquires an image from the camera 101 (S101).
  • the camera 101 generates a video of the scene
  • the video acquisition unit 110 acquires the video output from the camera 101 (input video).
  • the input video image includes a person working at the site and a work object such as a hammer used by the person.
  • the terminal 100 detects an object based on the acquired input video (S102).
  • the detection unit 120 uses an object recognition engine to detect a rectangular area in an image included in the input video, recognizes an object within the detected rectangular area, and assigns a label to the recognized object. For each detected object, the detection unit 120 outputs an object label and position information of a rectangular area of the object as an object detection result. For example, when object detection is performed from the image in FIG. 10, a person and a hammer are detected as shown in FIG. 11, and a rectangular area of the person and a rectangular area of the hammer are detected.
  • the terminal 100 determines a gaze area in the input video based on the object detection result (S103).
  • the first determination unit 131 of the image quality change determination unit 130 extracts an object having a label to be a gaze target based on the object detection result of each object.
  • the first determination unit 131 extracts objects whose object label is a person or a work object from the detected objects, and determines a rectangular area of the corresponding object as a gaze area.
  • a person and a hammer are detected in the image, and since the hammer corresponds to a work object, a rectangular area of the person and a rectangular area of the hammer are determined to be the gaze area.
  • the terminal 100 encodes the input video based on the determined gaze area (S104).
  • the compression efficiency determining unit 140 encodes the input video so that the region of interest has higher image quality than other regions.
  • the image quality of the person's rectangular area and the hammer's rectangular area is improved by lowering the compression ratio of the person's rectangular area and the hammer's rectangular area than the compression rate of other areas.
  • the terminal 100 transmits the encoded data to the center server 200 (S105), and the center server 200 receives the encoded data (S106).
  • Terminal communication unit 150 transmits encoded data with high image quality of the gaze area to base station 300.
  • the base station 300 transfers the received encoded data to the center server 200 via the core network or the Internet.
  • Center communication unit 210 receives the transferred encoded data from base station 300.
  • the center server 200 decodes the received encoded data (S107).
  • the decoder 220 decodes the encoded data according to the compression rate of each region, and generates a video (received video) in which the gaze region is of high quality.
  • the center server 200 recognizes the behavior of the object based on the decoded received video (S108).
  • the predictors PM1 to PM3 of the behavior recognition unit 230 each analyze the received video and recognize the behavior of the object.
  • FIG. 12 shows an example of behavior recognition processing by the predictor PM of the behavior recognition unit 230 shown in FIG. 8.
  • the object detection unit 231 first detects an object in the input received video (S201).
  • the object detection unit 231 uses an object recognition engine to detect a rectangular area in each image included in the received video, recognizes an object within the detected rectangular area, and assigns a label to the recognized object.
  • the object detection unit 231 outputs an object label and position information of a rectangular area of the object as an object detection result.
  • the tracking unit 232 tracks the detected object in the received video (S202).
  • the tracking unit 232 assigns a tracking ID to each detected object, and tracks the object identified by the tracking ID in each image.
  • the behavior predictor 233 predicts the behavior of each tracked object (S203).
  • the behavior predictor 233 uses a behavior recognition engine to predict a person's behavior from a video including a tracked person and a work object.
  • the behavior predictor 233 outputs the predicted behavior label and the score of each behavior label.
  • the behavior determination unit 234 determines the behavior of the object based on the score of the predicted behavior label (S204).
  • a person and a hammer are detected by tracking.
  • the behavior predictor 233 recognizes the behavior of the person based on the detected image of the person and the hammer, and outputs a score for each behavior label. For example, the score for pegging is 0.8, the score for heavy machinery work is 0.1, the score for unsafe behavior is 0.0, and the score for non-work is 0.1. Then, since the score for pegging is the highest, the behavior determination unit 234 determines that the person's action is pegging. The behavior determination unit 234 outputs the determined behavior and the score of the behavior.
  • the center server 200 extracts the gaze target based on the analysis information analyzed by the action recognition process (S109).
  • the gaze target analysis unit 250 sets the person whose behavior has been recognized as the gaze target, and if the recognition target includes a work object, the work object may also be included in the gaze target. For example, in the example of FIG. 13, since the work of driving a pile is recognized from the person and the hammer, the person and the hammer whose work has been recognized may be the objects of attention.
  • the gaze target analysis unit 250 outputs gaze target extraction information including position information of the extracted gaze target.
  • FIG. 14 shows an example of the operation of the gaze target extraction process according to this embodiment.
  • gaze target extraction processing is performed on the recognition results of each object.
  • the gaze target analysis unit 250 acquires the prediction results of a plurality of predictors, that is, the behavior recognition results (S301).
  • the gaze target analysis unit 250 obtains behavior recognition results including behavior labels and scores predicted by the predictors PM1 to PM3 of the behavior recognition unit 230.
  • the behavior label with the highest score output by the behavior determination unit 234 of each predictor is acquired.
  • a plurality of behavior labels and scores output from the behavior predictor 233 of each predictor may be acquired, or an arbitrary number of behavior labels output from the behavior predictor 233 of each predictor may be acquired. You may also obtain behavior labels. For example, the top three behavior labels with the highest scores may be acquired.
  • a gaze target is extracted based on the dispersion of the plurality of behavior labels of each predictor.
  • the gaze target analysis unit 250 determines the dispersion of the plurality of action recognition results (S302). For example, the gaze target analysis unit 250 determines variations in behavior labels included in the behavior recognition results of a plurality of predictors. The presence or absence of variation in behavior labels may be determined, or the magnitude of variation may be determined. Moreover, the variation may be determined not only by the variation in behavior labels but also by including the score of the behavior labels.
  • the gaze target analysis unit 250 determines the target object that performs the predicted action as the gaze target (S303). For example, if there are variations in the behavior labels, the gaze target analysis unit 250 determines the target object of the behavior label as the gaze target.
  • FIG. 15 shows an example in which it is determined that there is variation in recognition results when the predictors PM1 to PM3 have learned the same behavior. For example, the predictors PM1 to PM3 have learned the same task taken in different environments. In the example of FIG.
  • the behavior label of the prediction result of the predictor PM1 is heavy machinery work
  • the behavior label of the prediction result of the predictor PM2 is truck transportation
  • the behavior label of the prediction result of the predictor PM3 is compaction work
  • the gaze target analysis unit 250 may determine the target object of the behavior label as the gaze target. For example, if the number M of matching behavior labels with the highest scores among the N predictors is equal to or less than a threshold T, the predicted target object is determined to be the gaze target.
  • the behavior label of the prediction result of predictor PM1 is truck transport
  • the behavior label of the prediction result of predictor PM2 is compaction work
  • the behavior label of the prediction result of predictor PM3 is truck transportation
  • FIG. 16 shows an example in which it is determined that there are variations in recognition results when the predictors PM1 to PM3 have learned different behaviors.
  • the predictors PM1 to PM3 are learning different actions taken in the same environment that can be performed simultaneously.
  • the predictor PM1 is a predictor that has learned the posture (skeleton) of a person's behavior
  • the predictor PM2 is a predictor that has learned the content (work) of a person's behavior
  • the predictor PM3 is a predictor that has learned the posture (skeleton) of a person's behavior.
  • This is a predictor that has learned detailed information on inspection work among the actions recognized by device PM2. As shown in FIG.
  • the recognition results of the predictors PM1 and PM2 each include a plurality of behavior labels with scores of the same level or a predetermined value or higher, it is determined that the recognition results of the predictors PM1 to PM3 are dispersed, and the target Decide on an object as a gaze target.
  • the predictor that responds to the target object's behavior is limited to the one that has learned the target behavior. It is expected that For example, in an example where predictor A and predictor B are used, it is assumed that compaction work and heavy machinery work cannot be performed at the same time, and that predictor A is learning compaction work and predictor B is learning heavy machinery work. . In this example, when the worker is performing compaction work, it is expected that only predictor A will respond, the score for compaction work will be high, and the score of predictor B for heavy equipment work will be low.
  • the scores of both predictors become high, it can be determined that the actions are indistinguishable and the uncertainty of recognition is high. Therefore, it may be determined that the uncertainty is large depending on whether there are a plurality of behavior classes with scores equal to or higher than a certain score threshold between different predictors. For example, in the recognition result of predictor A, the score for compaction work is 0.8, in the recognition result of predictor B, the score for heavy equipment work is 0.9, and the score threshold is 0.6. Since there are two behavior classes that exceed the score threshold, it is determined that the uncertainty is large.
  • the gaze target analysis unit 250 excludes the target object that performs the predicted action from the gaze targets (S304). That is, in this case, the gaze target analysis unit 250 does not select the object as the gaze target. For example, if there is no variation in the behavior labels, the gaze target analysis unit 250 does not select the target object of the behavior label as the gaze target.
  • FIG. 17 shows an example in which it is determined that there is no variation in recognition results when the predictors PM1 to PM3 have learned the same behavior. Similar to FIG. 15, for example, the predictors PM1 to PM3 each learn the same task photographed in different environments.
  • the activity labels of the prediction results of the predictors PM1 to PM3 are all heavy machinery work, and since the prediction results match, it is determined that there is no variation, and the predicted target object is excluded from the gaze target.
  • the gaze target analysis unit 250 may exclude the target object of the behavior label from the gaze targets.
  • the prediction target is excluded from the gaze target.
  • the behavior label of the prediction result of the predictor PM1 is transporting a trolley
  • the behavior label of the prediction result of the predictor PM2 is transporting a trolley
  • the behavior label of the prediction result of the predictor PM3 is transporting a trolley
  • FIG. 18 shows an example in which it is determined that there is no variation in recognition results when the predictors PM1 to PM3 have learned different behaviors.
  • the predictors PM1 to PM3 learn different actions taken in the same environment that can be performed simultaneously.
  • the predictor PM1 learns the posture of a person's action
  • the predictor PM2 is a predictor that has learned the content (work) of a person's actions
  • the predictor PM3 is a predictor that has learned detailed information on inspection work among the actions that the predictor PM2 recognizes. This is a predictor that has learned .
  • FIG. 18 shows an example in which it is determined that there is no variation in recognition results when the predictors PM1 to PM3 have learned different behaviors.
  • the predictors PM1 to PM3 learn different actions taken in the same environment that can be performed simultaneously.
  • the predictor PM1 learns the posture of a person's action
  • the predictor PM2 is a predictor that has learned the content (work) of a person's actions
  • the center server 200 notifies the terminal 100 of the extracted gaze target information extracted by the gaze target extraction process (S110), and the terminal 100 acquires the extracted gaze target information. (S111).
  • the center communication unit 210 transmits extracted gaze target information indicating the position of the extracted gaze target to the base station 300 via the Internet or the core network.
  • the base station 300 transfers the received extracted gaze target information to the terminal 100.
  • Terminal communication unit 150 receives the transferred extracted gaze target information from base station 300.
  • the terminal 100 determines a gaze area based on the received extracted gaze target information (S112).
  • the second determination unit 132 of the image quality change determination unit 130 determines the area indicated by the extracted gaze target information notified from the center server 200 as the gaze area.
  • the extracted gaze target information indicates a rectangular area of a person and a rectangular area of a hammer, and these areas are determined as the gaze area.
  • a circumscribed area including a rectangular area of the person and a rectangular area of the hammer may be set as the gaze area. This circumscribed area may be notified from the center server 200 to the terminal 100. Thereafter, S104 to S112 are repeated.
  • a gaze target is extracted based on the indeterminacy of the behavior recognition result, and the image quality of the area containing the extracted gaze target is sharpened. do.
  • the results of action recognition vary, it is assumed that the action cannot be recognized correctly from the video. Therefore, by making a judgment based on the indeterminacy of the action recognition result, it is possible to appropriately select the object to be focused on. Therefore, depending on the action recognition result, it is possible to ensure the image quality of a specific part including the target to be watched, and to compress other areas, making it possible to reduce the amount of data to be transmitted and prevent mistakes in action recognition.
  • Embodiment 2 Next, a second embodiment will be described.
  • the behavior recognition unit outputs the uncertainty of the behavior recognition result.
  • the configuration other than the behavior recognition unit is the same as that in FIGS. 5 and 6 of Embodiment 1, so a description thereof will be omitted. Note that this embodiment can be implemented in combination with Embodiment 1, and each configuration shown in Embodiment 1 may be used as appropriate.
  • FIG. 20 shows a configuration example of the behavior recognition unit 230 according to this embodiment.
  • the behavior recognition unit 230 includes an object detection unit 231, a tracking unit 232, a behavior predictor 233, and a behavior determination unit 234 similar to the predictor PM in FIG. , further includes a dropout setting section 235 and a variation calculation section 236.
  • the dropout setting unit 235 sets dropout in the neural network of the behavior predictor 233.
  • the variation calculation unit 236 calculates the variation in the prediction results predicted multiple times by the behavior predictor 233 that has set dropout, that is, the recognition results recognized multiple times.
  • FIG. 21 shows an example of behavior recognition processing by the behavior recognition unit 230 shown in FIG. 20. Note that the other operations are similar to those in FIG. 9 of the first embodiment.
  • the object detection unit 231 detects an object in the input received video (S201), and the tracking unit 232 detects the object in the detected received video. (S202).
  • the dropout setting unit 235 sets dropout in the behavior predictor 233 (S211).
  • the neural network of the behavior predictor 233 includes an input layer, a hidden layer (middle layer), and an output layer.
  • the input layer includes multiple nodes and the hidden layer includes multiple nodes.
  • the dropout setting unit 235 selects, for example, a node in the hidden layer and inactivates the selected node. Randomly selected nodes may be inactivated, or nodes may be selected and inactivated so as to achieve a predetermined dropout rate. Note that not only nodes in the hidden layer but also nodes in the input layer may be inactivated.
  • the behavior predictor 233 predicts the behavior of the object as in the first embodiment (S203), and the behavior determination unit 234 determines the behavior label of the predicted behavior.
  • the behavior of the object is determined based on the score (S204).
  • the behavior determination unit 234 outputs the determined behavior recognition result.
  • S211, S203, and S204 are repeated multiple times, and action recognition by dropout is performed multiple times. Recognition may be performed multiple times by sequentially repeating the processing in S211, S203, and S204, or may be performed in parallel by duplicating the series of processing in S211, S203, and S204. However, in multiple recognitions, the nodes to be inactivated in S211 are set differently each time.
  • the variation calculation unit 236 calculates the variation in the plurality of action recognition results obtained by predicting the action multiple times (S212). Similar to the first embodiment, the variation calculation unit 236 may calculate the variation in the behavior labels included in the behavior recognition results, or may calculate the variation in the scores of the behavior labels. The variation calculation unit 236 may determine whether there is variation in the behavior labels and output the determined result. If the behavior labels of the plurality of behavior recognition results are different, "with variation” is output as the variation calculation result, and when the behavior labels of the plurality of behavior recognition results match, "no variation” is outputted as the variation calculation result.
  • the variation calculation unit 236 may determine whether the variation in behavior labels of the plurality of behavior recognition results is larger than a threshold value, and output the determined result. As in Embodiment 1, for example, if the number M of matching behavior labels with the highest score among N times of behavior recognition (inference) is less than or equal to the threshold value T, "with variation” is output as the variation calculation result. However, if the number M of matching behavior labels with the highest score is greater than the threshold T, no variation is output as the variation calculation result. The degree of variation, which is the ratio of the number M of matching behavior labels with the highest score to the N times of behavior recognition, may be output as the variation calculation result. Note that, similarly to the first embodiment, the gaze target analysis unit 250 may calculate and determine the variation.
  • the gaze target analysis unit 250 extracts the gaze target based on the variation calculation result calculated by the variation calculation unit 236 of the behavior recognition unit 230. For example, when the presence or absence of variation is output as the variation calculation result, if there is variation, the target object of the action recognition result is determined as the gaze target. When the degree of dispersion is output as the dispersion calculation result, the gaze target may be determined according to the comparison result between the degree of dispersion and the threshold value.
  • a gaze target may be extracted using a single behavior predictor that outputs the uncertainty of behavior prediction. Even in this case, as in Embodiment 1, since the gaze target can be appropriately selected, it is possible to prevent behavior recognition mistakes while suppressing the amount of data to be transmitted. Furthermore, the uncertainty of recognition results can be determined without preparing multiple predictors.
  • the center server extracts the gaze target and the terminal determines the gaze area based on the extracted gaze target, but the center server determines the gaze area based on the extracted gaze target. It's okay.
  • the center server may notify the terminal of the coordinates of the gaze area and the size of the area.
  • processing flow described in the above embodiment is an example, and the order of each process is not limited to the above example.
  • the order of some of the processes may be changed, or some of the processes may be executed in parallel.
  • Each configuration in the embodiments described above is configured by hardware, software, or both, and may be configured from one piece of hardware or software, or from multiple pieces of hardware or software.
  • Each device and each function (processing) may be realized by a computer 30 having a processor 31 such as a CPU (Central Processing Unit) and a memory 32 as a storage device, as shown in FIG.
  • a program for performing the method (video processing method) in the embodiment may be stored in the memory 32, and each function may be realized by having the processor 31 execute the program stored in the memory 32.
  • These programs include instructions (or software code) that, when loaded into a computer, cause the computer to perform one or more of the functions described in the embodiments.
  • the program may be stored on a non-transitory computer readable medium or a tangible storage medium.
  • computer readable or tangible storage media may include random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD - Including ROM, digital versatile disc (DVD), Blu-ray disc or other optical disc storage, magnetic cassette, magnetic tape, magnetic disc storage or other magnetic storage device.
  • the program may be transmitted on a transitory computer-readable medium or a communication medium.
  • transitory computer-readable or communication media includes electrical, optical, acoustic, or other forms of propagating signals.
  • an image quality control means for controlling the image quality of a gaze area including a gaze target in an input video
  • recognition means that performs recognition processing regarding an object included in the video whose image quality in the gaze area is controlled
  • Extracting means for extracting the gaze target based on the uncertainty of the recognition result of the recognition process
  • a video processing system equipped with The recognition process includes a process of recognizing the behavior of the object, The extraction means extracts the gaze target based on variations in the recognition results of the behavior.
  • the extracting means determines the object for which the action has been recognized as the gaze target if the variation in the recognition result of the action is not within a predetermined range.
  • the video processing system described in Appendix 2. includes a plurality of behavior predictors that have learned different learning data, The extraction means extracts the gaze target based on variations in the recognition results of the plurality of actions output from the plurality of action predictors.
  • the video processing system according to appendix 2 or 3. Appendix 5)
  • the recognition means includes a behavior predictor, The extraction means extracts the gaze target based on variations in recognition results of a plurality of actions outputted when the action predictor performs action recognition a plurality of times.
  • the video processing system according to appendix 2 or 3.
  • the extraction means extracts the gaze target based on variations in behavior labels included in the recognition results of the behaviors or variations in scores of the behavior labels.
  • the video processing system according to any one of Supplementary Notes 2 to 5.
  • the behavior predictor outputs recognition results of the plurality of behaviors by inactivating different nodes of the neural network each time the behavior is recognized.
  • the video processing system according to appendix 5.
  • (Appendix 8) Controls the image quality of the gaze area including the gaze target in the input video, performing recognition processing on an object included in the video whose image quality in the gaze area is controlled; extracting the gaze target based on the uncertainty of the recognition result of the recognition process; Video processing method.
  • the recognition process includes a process of recognizing the behavior of the object, extracting the gaze target based on variations in the recognition results of the behavior;
  • the video processing method described in Appendix 8. (Appendix 10) In the extraction of the gaze target, if the variation in the recognition result of the behavior is not included in a predetermined range, the object for which the behavior has been recognized is determined as the gaze target;
  • the video processing method according to appendix 9. (Appendix 11) Recognizing the behavior of the object using multiple behavior predictors that have learned different learning data, extracting the gaze target based on variations in the recognition results of the plurality of actions output from the plurality of action predictors;
  • (Appendix 12) recognizing the behavior of the object by a behavior predictor; extracting the gaze target based on variations in recognition results of a plurality of actions output by the action predictor performing action recognition multiple times; The video processing method according to appendix 9 or 10.
  • (Appendix 13) extracting the gaze target based on variations in behavior labels included in the behavior recognition results or variations in scores of the behavior labels; The video processing method according to any one of Supplementary Notes 9 to 12.
  • the behavior predictor outputs recognition results of the plurality of behaviors by inactivating different nodes of the neural network each time the behavior is recognized.
  • an image quality control means for controlling the image quality of a gaze area including a gaze target in an input video; recognition means that performs recognition processing regarding an object included in the video whose image quality in the gaze area is controlled; Extracting means for extracting the gaze target based on the uncertainty of the recognition result of the recognition process;
  • An image processing device comprising: (Appendix 16) The recognition process includes a process of recognizing the behavior of the object, The extraction means extracts the gaze target based on variations in the recognition results of the behavior. The video processing device according to appendix 15. (Appendix 17) The extracting means determines the object for which the action has been recognized as the gaze target if the variation in the recognition result of the action is not within a predetermined range.
  • the video processing device includes a plurality of behavior predictors that have learned different learning data, The extraction means extracts the gaze target based on variations in the recognition results of the plurality of actions output from the plurality of action predictors.
  • the video processing device according to appendix 16 or 17.
  • the recognition means includes a behavior predictor, The extraction means extracts the gaze target based on variations in recognition results of a plurality of actions outputted when the action predictor performs action recognition a plurality of times.
  • the video processing device according to appendix 16 or 17.
  • the behavior predictor outputs recognition results of the plurality of behaviors by inactivating different nodes of the neural network each time the behavior is recognized.
  • the video processing device according to appendix 19.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Analysis (AREA)

Abstract

Un système de traitement de vidéo (10) comprend : une unité de commande de qualité d'image (11) qui commande la qualité d'image dans une région du regard incluant un sujet de regard dans une vidéo qui est entrée dans le système de traitement de vidéo (10) ; une unité de reconnaissance (12) qui réalise un processus de reconnaissance relatif à un objet qui est inclus dans la vidéo dont la qualité d'image dans la région du regard est commandée par l'unité de commande de qualité d'image (11) ; et une unité d'extraction (13) qui, sur la base de l'incertitude des résultats de reconnaissance provenant du processus de reconnaissance réalisé par l'unité de reconnaissance (12), extrait le sujet de regard inclus dans la région du regard que l'unité de commande de qualité d'image (11) commande.
PCT/JP2022/032586 2022-08-30 2022-08-30 Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo WO2024047748A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/JP2022/032586 WO2024047748A1 (fr) 2022-08-30 2022-08-30 Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2022/032586 WO2024047748A1 (fr) 2022-08-30 2022-08-30 Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo

Publications (1)

Publication Number Publication Date
WO2024047748A1 true WO2024047748A1 (fr) 2024-03-07

Family

ID=90098933

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2022/032586 WO2024047748A1 (fr) 2022-08-30 2022-08-30 Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo

Country Status (1)

Country Link
WO (1) WO2024047748A1 (fr)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001145101A (ja) * 1999-11-12 2001-05-25 Mega Chips Corp 人物画像圧縮装置
JP2020068521A (ja) * 2018-10-19 2020-04-30 ソニー株式会社 センサ装置、信号処理方法
JP2021149446A (ja) * 2020-03-18 2021-09-27 株式会社日立製作所 注視物体認識システム及び方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001145101A (ja) * 1999-11-12 2001-05-25 Mega Chips Corp 人物画像圧縮装置
JP2020068521A (ja) * 2018-10-19 2020-04-30 ソニー株式会社 センサ装置、信号処理方法
JP2021149446A (ja) * 2020-03-18 2021-09-27 株式会社日立製作所 注視物体認識システム及び方法

Similar Documents

Publication Publication Date Title
EP3735625B1 (fr) Procede et dispositif pour entrainer le navigateur d'un robot de suivi d'objets
US10748061B2 (en) Simultaneous localization and mapping with reinforcement learning
US11748896B2 (en) Object tracking method and apparatus, storage medium, and electronic device
US20170344832A1 (en) System and method for event monitoring and detection
KR102330055B1 (ko) 드론을 이용한 구조물의 변화 검출 방법 및 시스템
CN111988524A (zh) 一种无人机与摄像头协同避障方法、服务器及存储介质
JP2022548009A (ja) 物体移動システム
CN113901911B (zh) 图像识别、模型训练方法、装置、电子设备及存储介质
CN113505704B (zh) 图像识别的人员安全侦测方法、系统、设备及存储介质
CN116866520B (zh) 一种基于ai的单轨吊安全运行实时监控管理系统
WO2024047748A1 (fr) Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo
CN113095160A (zh) 基于人工智能和5g的电力系统人员安全行为识别方法及系统
JP2022526071A (ja) 状況認識監視
WO2024047791A1 (fr) Système de traitement vidéo, procédé de traitement vidéo et dispositif de traitement vidéo
WO2024042705A1 (fr) Système de traitement vidéo, procédé de traitement vidéo et dispositif de traitement vidéo
WO2024047747A1 (fr) Système de traitement de vidéo, procédé de traitement de vidéo et dispositif de traitement de vidéo
WO2024047794A1 (fr) Système de traitement de vidéo, dispositif de traitement de vidéo et procédé de traitement de vidéo
WO2024047790A1 (fr) Système de traitement vidéo, dispositif de traitement vidéo et procédé de traitement vidéo
WO2024038517A1 (fr) Système de traitement vidéo, procédé de traitement vidéo et dispositif de commande d'image
CN111695404B (zh) 行人跌倒检测方法、装置、电子设备及存储介质
WO2024047793A1 (fr) Système de traitement vidéo, dispositif de traitement vidéo et procédé de traitement vidéo
KR102138006B1 (ko) 보안 로봇 제어 방법 및 그 방법을 지원하는 장치
CN113128414A (zh) 人员跟踪方法、装置、计算机可读存储介质及电子设备
WO2024057446A1 (fr) Système de traitement vidéo, dispositif de traitement vidéo et procédé de traitement vidéo
KR102647135B1 (ko) 인공지능 기반 객체 탐지 알고리즘을 이용하는 시공 현장을 위한 실시간 균열 탐지 시스템 및 이를 위한 동작 방법

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22957351

Country of ref document: EP

Kind code of ref document: A1